Objectives
Target trial emulation is a framework for conducting causal inference using observational data. We employ this framework to estimate the effect of changing planned participation duration, measured using the acute:chronic workload ratio (ACWR), on injury risk among adolescent ice hockey players without recent injuries.
Design
Prospective cohort study designed to emulate a hypothetical randomized trial.
Methods
We used data from a 5-year cohort study (2013–2018) of ice hockey players aged 13–17 years in Alberta and British Columbia. We estimated injury risks associated with different planned changes in hockey participation duration (e.g. half [ACWR = 0.5], no change [ACWR = 1], two-fold [ACWR = 2], three-fold [ACWR = 3], and five-fold [ACWR = 5]) relative to participation in the previous 4 weeks. Outcomes were modeled using generalized additive models. We conducted secondary analyses restricted to concussions, and stratified by league bodychecking status.
Results
There were 2633 eligible participants, contributing 115,821 player-trials. Injury risk was 1.9 % (95 % CI: 1.7 %–2.3 %) for no change in participation (ACWR = 1). Injury risk ratios (RRs) were 0.43 at ACWR = 0.5 (95 % CI: 0.31–0.54), 1.62 (95 % CI: 1.33–1.98) at ACWR = 2, 1.91 at ACWR = 3 (95 % CI: 1.52–2.48) and 2.35 at ACWR = 5 (95 % CI: 1.68–3.26). Patterns were similar by league bodychecking status. Concussion RRs were stable between ACWR = 1 and 1.5, but RRs were greater than for any injury past ACWR = 2.
Conclusions
Within the assumptions of this target trial emulation, injury risk increases consistently (no sweet spots) for increases in planned changes in participation duration relative to the previous 4 weeks among adolescent ice hockey players without recent injuries. Injuries in injury risk are less than expected for the increased exposure time at risk, suggesting beneficial effects of increasing participation that partially counteract the increased exposure time.